Most modern cameras have a feature to reduce noise in long exposures. The process normally involves the camera taking a second exposure with the shutter closed to get a base noise profile from the sensor to subtract from the previously taken shot.
How effective is it?
To demonstrate how the camera performs this operation I used the following workflow:
- I took two long exposures using a 6D at 3200ISO, Body cap on, view finder closed, all NR off.
- Each exposure was 430 seconds.
- Using Photoshop I imported both shots with +2EV, then overlaid the two images on separate layers.
- I then inverted (Command I) the top layer and reduced it’s transparency to 50%.
If both images are identical the result will be a perfect uniform 50% grey image. However both images from the camera are never perfectly the same and you can see the resulting variations in the grey square representing the noise that long exposure NR will not be able to correct.
Here is an example using the top left 300×300 pixels from the shots.
The algorithm used by the camera will be slightly smarter and it would only subtract distinct hot pixels at a certain threshold or criteria, but you get the idea.
Here is the same exposure with long exposure NR enabled in the camera, you can see that the resulting image has about the same amount of noise as the example above.
For those of us shooting long exposures it’s important to get the best sensor we can to start with, as in camera NR does not remove all the noise, and you have the overhead of having to waiting the exposure length a second time while the camera creates a sensor profile for subtraction.
Of course there are many post production tools to remove noise, but it always helps to start with as little as possible.
Check out my sensor comparison page for how long exposures compare across some popular models.